The Machine That Splits Water: Photosystem II and the Origin of Oxygen

The Hook: The breath you are taking right now comes from a torn-apart water molecule

Take a deep breath. About a fifth of the air now flowing into your lungs is molecular oxygen. Every single one of those O₂ molecules was produced somewhere on Earth by a machine smaller than a millionth of a millimeter, a machine with exactly one job: to use the energy of sunlight to tear water molecules apart.

That sounds more harmless than it is. Splitting water – separating the oxygen atoms from the hydrogen atoms and forming an O–O bond between two oxygen atoms – is among the most chemically demanding reactions that occur anywhere in living nature. Water is extraordinarily stable; we use it to put out fires precisely because it is so reluctant to react. To oxidize a water molecule you must wrest electrons away from it, and water holds its electrons very tightly. In the laboratory, humans still manage this only with considerable energy input, high voltages, and expensive precious-metal catalysts such as platinum or iridium.

Nature has been doing the same thing for roughly three billion years – at room temperature, in almost every green leaf, with a catalyst made of four manganese atoms, one calcium atom, and a few oxygen bridges. This tiny cluster of metal atoms, barely larger than a nanometer, sits at the heart of a protein complex called Photosystem II, and it is arguably the most consequential machine evolution ever built. Because its waste product is oxygen – the oxygen of the oceans, of the atmosphere, of your cells. Without this machine there would be no breathing animals, no ozone layer, no humans.

This article tells the story of that molecular water-splitting machine: how researchers spent two centuries figuring out that plants "purify" the air and where the released oxygen really comes from; how it turned out that photosynthesis requires not one but two light reactions wired in series; how an elegant flash-of-light experiment revealed nature's hidden "oxygen clock"; and how, in recent years, state-of-the-art X-ray lasers have for the first time been able to film an oxygen molecule literally coming into being on four manganese atoms.

Part 1: Two centuries of detective work

Priestley and the "restored" air

The trail begins in 1771 with the English polymath Joseph Priestley. He noticed that a candle in a sealed glass vessel soon goes out and a mouse inside it suffocates – the air had been "spoiled." But when he placed a sprig of mint in the jar as well, the air became "good" again after a few days: the candle would burn once more, and the mouse survived. Without knowing it, Priestley had discovered oxygen production by plants. What escaped him: the effect occurred only in the light.

This decisive role of light was recognized a few years later by the Dutch physician Jan Ingenhousz (1779). In meticulous experiments he showed that only the green parts of a plant improve the air, and only when sunlight falls on them; in darkness the process even reverses (the plant's respiration). Around 1800 the Genevan naturalist Nicolas-Théodore de Saussure completed the material balance: the plant takes up carbon dioxide and water and builds its substance from them. With that, the overall equation of photosynthesis was established in outline, long before anyone knew its molecules:

Carbon dioxide + water + light energy → sugar + oxygen

In modern notation, simplified:

6 CO₂ + 6 H₂O + light → C₆H₁₂O₆ + 6 O₂

Another beautiful piece of the mosaic came from the German botanist Theodor Wilhelm Engelmann (1882). He placed a filamentous green-algal strand under the microscope and illuminated it with a spectrum of dispersed sunlight. Then he added oxygen-loving bacteria and watched where they gathered – precisely where the alga released the most oxygen. In this way, using living bacteria as oxygen detectors, he measured the first action spectrum of photosynthesis: strongest in red and blue light, weak in the green. This explains, incidentally, why leaves look green – they reflect the green light they can use the least.

Where does the oxygen come from – the CO₂ or the water?

One puzzle remained open for a long time: the released oxygen – does it come from the absorbed carbon dioxide (CO₂) or from the water (H₂O)? Both, after all, contain oxygen atoms. The intuitive answer of most researchers was: from the CO₂.

The intuitive answer was wrong. The proof came in 1941 from the American chemists Samuel Ruben and Martin Kamen in a classic experiment. They fed algae water in which the oxygen was labeled with the rare, heavy isotope ¹⁸O, together with ordinary CO₂ – and then the other way around. The result was unambiguous: the released O₂ always carried the isotopic signature of the water, never that of the CO₂. The oxygen we breathe therefore comes from split water. The CO₂ takes an entirely different route – it is built into sugar later.

With that, the true chemical feat of photosynthesis had been named: water oxidation, or water splitting. In formula form:

2 H₂O → O₂ + 4 H⁺ + 4 e⁻

Four electrons and four protons must be wrested from the water, and two oxygen atoms joined into O₂. That is the hard part. Everything else in photosynthesis – the building of sugar – is chemically comparatively routine.

The Hill reaction: water splitting in isolation

An important methodological breakthrough was achieved by the British biochemist Robert Hill in 1937. He isolated chloroplasts from leaves, gave them no CO₂ but instead an artificial electron acceptor (an iron salt) – and, lo and behold, when illuminated they produced oxygen, with no sugar building at all. The Hill reaction proved that water splitting and oxygen release are an independent process, separable from the CO₂. This made the "light reaction" experimentally tangible – it could be studied in a test tube.

Part 2: Not one but two light machines

The Emerson effect and the puzzling "red drop"

In the 1940s and 1950s the American biophysicist Robert Emerson studied the efficiency of photosynthesis at different colors of light in detail. In doing so he stumbled upon a peculiarity. When he illuminated algae with deep-red light (wavelengths beyond about 680 nanometers), the efficiency suddenly collapsed – the so-called red drop. Strange, because chlorophyll certainly does absorb that light.

Even stranger was Emerson's second observation (around 1957): if he simultaneously added some shorter-wavelength red light (around 650 nm), the efficiency shot up – and by more than the sum of the two light sources individually would suggest. Two colors of light together thus produced more than both separately. This Emerson enhancement effect was a puzzle – and at the same time the decisive clue.

The Z-scheme: two photosystems in series

The interpretation was provided by Robin Hill and Fay Bendall in 1960. Their bold thesis: photosynthesis works not with one light-driven reaction but with two photosystems wired in series, optimized for slightly different wavelengths and working together. The Emerson effect then explained itself: deep-red light alone can drive only one of the systems; only when both systems receive light at the same time does the chain run smoothly.

The two machines were named – somewhat counterintuitively, in the order of their discovery rather than of their operation – Photosystem I (absorption maximum at ~700 nm, hence its reaction center "P700") and Photosystem II (maximum at ~680 nm, reaction center "P680"). The flow of electrons runs, contrary to the numbering, from II to I.

If you plot the energy levels of the electrons along their path through the chain, you get a characteristic zigzag line that gave the whole thing its name: the Z-scheme. The sequence in brief:

1. Photosystem II catches a quantum of light. The energy lifts an electron in the chlorophyll P680 to a high level; the electron is released and begins its journey. The "hole" left behind is extremely electron-hungry – P680⁺ is the strongest oxidant biology knows.
2. This hole is filled by PSII wresting an electron from water. Right here – at the filling of the hole – the water is split and the oxygen is released.
3. The released electron travels downhill via a chain of carriers (plastoquinone, the cytochrome b₆f complex, plastocyanin), pumping protons across the membrane as it goes. This proton gradient then drives the ATP synthase – the same rotary machine that also makes ATP in our mitochondria (see cross-references).
4. Photosystem I catches another quantum of light and lifts the now-descended electron to a high level once more – from there it ultimately lands on the molecule NADP⁺, which becomes NADPH.

At the end of the light reaction stand two energy-rich products: ATP (the energy currency) and NADPH (an electron carrier). Both are then used in the Calvin-Benson cycle to build sugar from CO₂. This carbon-fixation pathway was deciphered by Melvin Calvin, Andrew Benson, and James Bassham in the late 1940s and the 1950s using the radioactive carbon isotope ¹⁴C; Calvin received the Nobel Prize in Chemistry for it in 1961. The central (and notoriously inefficient) enzyme of this cycle, RuBisCO, is probably the most abundant protein on Earth.

Important for our topic: sugar building is the "second half." The actual water splitting – the origin of all oxygen – takes place exclusively in Photosystem II, in step 2 above. It is to that step that we now turn in detail.

Part 3: Kok's oxygen clock – the machine counts to four

A flash-of-light experiment with a surprising pattern

How does Photosystem II take water apart? Releasing oxygen is impossible in a single step, because the splitting requires the removal of four electrons (see the equation above), while a single photon supplies only the energy to transfer one electron. There must therefore be a mechanism that collects four separate light events before oxygen is produced in one stroke.

The experimental proof of this was provided in 1969 by the French researcher Pierre Joliot with an elegant idea. He kept algae in the dark so that their photosystems settled into a resting state, and then fired a sequence of very short, saturating flashes of light at them – each flash short enough to trigger every reaction center exactly once. Then he measured how much oxygen was released after each individual flash.

The result was astonishing. The oxygen did not come out evenly but in a clear pattern with a period of four: after the first and second flashes almost no oxygen appeared; the first large burst came after the third flash, the next after the seventh, then after the eleventh, and so on – a maximum every four flashes. With each cycle the pattern blurred a little more, until after many flashes it settled into a constant average.

Bessel Kok's model of the five S-states

The interpretation was supplied in 1970 by the Dutch-American biophysicist Bessel Kok. He proposed that the oxygen-evolving machine – the Oxygen-Evolving Complex (OEC) – is a kind of charge accumulator that passes through five states, which he called S₀ to S₄. Each flash of light wrests exactly one electron from the machine and advances it one step: S₀ → S₁ → S₂ → S₃ → S₄. Only in state S₄, once four positive charges have been accumulated, does the actual chemistry occur spontaneously: two water molecules are joined into one O₂, the oxygen is released, and the machine jumps back to S₀. This model is known to this day as the Kok clock or the S-state cycle model.

It explained the puzzling pattern at a stroke – including the seemingly strange maximum at the third, not the fourth, flash. Kok's solution: in the dark the machine does not rest in S₀ but predominantly in S₁. S₁ is the stable dark resting state. So if the machine starts at S₁, it needs only three more flashes to reach S₄ – hence the first oxygen after flash number three. And the reason the pattern blurs over time is that not every flash hits every center: some "miss" a step (misses), others accidentally trigger two (double hits). These small error rates add up and cause the initially sharp four-step waves to gradually dissolve into a smooth line.

The following overview summarizes the cycle:

Kok had deciphered the hidden logic of water splitting without ever having seen the machine. He knew that it counts to four – but not what it looks like. That answer would be decades in coming.

Part 4: The Mn4CaO5 cluster – a catalyst made of just a few atoms

The structure becomes visible

Only in the 2000s, with ever-better X-ray crystallography, could the Oxygen-Evolving Complex be resolved to the atomic scale. A milestone was set in 2011 by the Japanese researchers Yasufumi Umena, Keisuke Kawakami, Jian-Ren Shen and colleagues with a structure at 1.9 ångström resolution. What emerged was a tiny inorganic assembly at the heart of the enormous protein complex: a cluster of four manganese atoms, one calcium atom, and five bridging oxygen atoms – the molecular formula Mn₄CaO₅. Because of its shape, the cluster is often compared to a "distorted chair": a cube-like core of three manganese atoms and one calcium atom, with a fourth manganese atom hung "outside."

That is remarkably little. The entire world-shaping chemistry of biological oxygen production plays out on a catalyst made of a handful of atoms – surrounded and held precisely in position by a carefully folded protein scaffold and a network of hydrogen bonds and water channels. The manganese plays the role of charge accumulator: it can adopt several oxidation states (essentially Mn(III) and Mn(IV)) and thus take up, one after another, the four electrons wrested from the water.

The electron relay race: the tyrosine bridge

Between the light-driven chlorophyll P680 and the manganese cluster sits a decisive intermediary: a special amino-acid residue, the redox-active tyrosine Y_Z (tyrosine 161 of the protein subunit D1). When P680 loses its electron to the light and becomes the extremely "hungry" P680⁺, it first snatches its replacement electron from Y_Z. The now electron-poor tyrosine in turn pulls an electron from the manganese cluster. In this way the "hole" migrates step by step from the light to the water: P680⁺ → Y_Z → manganese cluster → water. In this way Photosystem II translates the fast, single one-electron excitations of light into the slow, bundled four-electron chemistry of water splitting.

Part 5: The film of O₂ formation – X-ray lasers in action

Why this was so hard to observe

Knowing the static structure is one thing; seeing how the cluster changes during the cycle is quite another. Two problems stood in the way. First, manganese is extremely sensitive to X-rays – the intense beams needed for structure determination reduce and damage the cluster before it can be imaged in its true state. Second, the transitions between the S-states are lightning-fast and occur at room temperature.

The solution came with a new class of instruments: the X-ray free-electron lasers (XFEL). These kilometer-long facilities produce unimaginably bright but extremely short X-ray pulses (on the order of femtoseconds, that is, millionths of a billionth of a second). The principle is called "diffraction before destruction": the pulse passes through the protein crystal and delivers its diffraction pattern before the beam destroys the molecule. With the method of serial femtosecond crystallography, a fine stream of tiny crystals is shot through the X-ray beam and flashed beforehand with laser light to set them into defined S-states. In this way the cluster can be, as it were, frozen – at room temperature, undamaged, at precisely chosen moments after one, two, or three flashes of light.

This technique was progressively refined from 2014 onward (Kern, Yano, Yachandra and colleagues, Nature). It allows the OEC to be filmed at work, so to speak – snapshot by snapshot.

What the snapshots showed (2023–2024)

In 2023 and 2024 this yielded spectacular insights, published among others in Nature and Nature Communications. Two findings stand out:

First, the researchers found that in the transition from S₂ to S₃ an additional oxygen atom (often called "O6") is incorporated into the cluster – as a bridge between the calcium atom and one of the manganese atoms (Mn1). This additional oxygen atom comes from an incoming water molecule and is the likely partner for the O–O bond being formed. With that, one of the oldest puzzles became tangible: where exactly in the cluster the second oxygen atom needed for O₂ comes from.

Second, they succeeded in illuminating the final and hardest-to-capture step – the transition S₃ → [S₄] → S₀, in which the oxygen is actually formed. The snapshots reveal a complex sequence over micro- to milliseconds: about 700 microseconds after the third flash, the incorporated bridging oxygen atom begins to change, in parallel with the reduction of the tyrosine Y_Z; the actual onset of O₂ release falls at about 1,200 microseconds. In this window the two oxygen atoms form their bond, the finished O₂ detaches, protons are released in a controlled manner through the hydrogen-bonding network, and the cluster returns to its starting state S₀. A paper in Nature Communications (2024) was devoted specifically to closing the Kok cycle – the question of how the resting state S₀ is restored after oxygen release.

It is remarkable: a model that Bessel Kok inferred in 1970 from a pattern of oxygen flashes – purely from the behavior of the machine, without seeing it – is today confirmed atom by atom, microsecond by microsecond, and filled with structure. This is science at its most beautiful: a bold hypothesis, confirmed in detail half a century later by entirely different methods.

I am of the opinion that, despite these breakthroughs, the exact reaction pathway of O–O bond formation is not yet definitively settled: several mechanistic models remain in competition (for instance, a nucleophilic attack by a water molecule on a highly oxidized manganese-oxo center versus the coupling of two bridging oxygen atoms). The most recent structural data strongly constrain the possibilities, but a scientific endpoint has not yet been reached.

Part 6: Why all of this matters – the oxygen of the world

The greatest pollution event in Earth's history

Deploy this molecular machine billions of times over billions of years, and you change an entire planet. The capacity for water splitting – oxygenic photosynthesis – arose in the cyanobacteria, probably around three billion years ago. Their waste product, oxygen, was a deadly poison to the living world of the time, which was adapted to oxygen-free conditions.

At first the released oxygen was chemically captured – above all by the iron dissolved in seawater, which oxidized to insoluble rust and sank to the seafloor. The enormous banded iron formations, from which we obtain most of our iron today, are the fossilized record of that era. Only once these sinks were saturated did oxygen begin to accumulate in the atmosphere – the so-called Great Oxidation Event around 2.4 billion years ago. It was perhaps the greatest environmental catastrophe and at the same time the greatest opportunity in Earth's history: it wiped out countless anaerobic life forms, but created the precondition for energy-rich oxygen respiration – and thus, ultimately, for all complex, multicellular life, ourselves included.

The cyanobacteria themselves were later, as recounted in the article on endosymbiotic theory (see cross-references), swallowed by another cell and became the chloroplasts of plants and algae. The Photosystem II in every blade of grass is therefore, in a direct line of descent, the very same invention that once transformed Earth's atmosphere. When you look at a leaf, you are seeing a three-billion-year-old, almost unchanged technology in operation.

A model for the energy transition?

Water splitting fascinates for a very present-day reason as well. Anyone who can split water into oxygen and hydrogen using sunlight holds a key to green hydrogen – a climate-neutral energy carrier. Technical water splitting (electrolysis) works but has so far often required expensive, rare precious metals as catalysts. Photosystem II accomplishes the same task with manganese and calcium – two of the most abundant and cheapest metals in the Earth's crust – at room temperature and with a high turnover rate.

No wonder, then, that the Mn₄CaO₅ cluster is a model for research into artificial photosynthesis. Chemists are trying to build manganese- and cobalt-based catalysts that match the robustness and efficiency of the natural template. I am of the opinion that an exact technical copy of the OEC has not yet been achieved – the biological machine repairs itself and is embedded in a highly complex protein environment that cannot simply be reproduced. But every detail the X-ray lasers reveal about the natural mechanism narrows the search space for better synthetic catalysts. Basic research on an algal leaf could thus contribute quite directly to the energy supply of tomorrow.

The Central Takeaway

Perhaps the most important lesson of Photosystem II is one about the art of breaking a hard problem into manageable sub-steps. Nature faces a task that is unsolvable in one go: it has to wrest four electrons from an extremely stable molecule all at once, yet it has only the energy for a single one per quantum of light. Its solution is not a feat of brute force but a charge accumulator that patiently collects one electron after another (the S-states) until the actual difficult reaction can run in a single, well-prepared moment.

This pattern – not forcing a hard event head-on, but preparing for it by incrementally accumulating state – is ubiquitous in engineering too. Anyone who builds systems knows it: you buffer events, gather state in small, low-loss steps, and trigger the expensive operation only when enough "charge" has been amassed. Nature's water-splitting machine, seen from this angle, is a three-billion-year-old lesson in elegant state management – with the incidental side effect of having created the atmosphere we breathe.

And it is a prime example of how science works: a pattern of oxygen flashes (Joliot) led to an abstract counting model (Kok), which half a century later was confirmed atom by atom and filled with life by X-ray lasers. Behavior first, structure later – the same methodological order we already encountered in the decipherment of Linear B and in the scaling laws of AI.

A Question to Reflect On

Photosystem II solves a seemingly impossible problem by patiently accumulating state instead of forcing power – and it stabilizes its cycle via a resting state (S₁) that buffers errors. Where in your own work are you trying to force a "four-electron reaction" in a single step, where a patient charge accumulator with well-defined intermediate states would be more robust and more elegant?

Cross-References in the Vault

• The Molecular Turbine: ATP Synthase and the Engine of Life – The ATP synthase, which converts the proton gradient of the light reaction into chemical energy: the same rotary machine that also drives our mitochondria.
• The Enemy That Became a Power Plant: Endosymbiosis and the Bacterial Origin of Complex Life – How the oxygen-producing cyanobacteria became the chloroplasts of plants, and thereby brought Photosystem II into every green cell.
• The Shape of Life: Levinthal's Paradox, Chaperones, and How Proteins Find Their Form – Why the exact folding of the protein scaffold that holds the Mn₄CaO₅ cluster in position is decisive.

Sources

• Nature (2023): Oxygen-evolving photosystem II structures during S₁–S₂–S₃ transitions – https://www.nature.com/articles/s41586-023-06987-5
• Nature (2023): Structural evidence for intermediates during O₂ formation in photosystem II – https://www.nature.com/articles/s41586-023-06038-z
• Nature Communications (2024): Closing Kok's cycle of nature's water oxidation catalysis – https://www.nature.com/articles/s41467-024-50210-6
• PMC (2024): Structure Function Studies of Photosystem II Using X-Ray Free Electron Lasers – https://pmc.ncbi.nlm.nih.gov/articles/PMC11321711/
• Biology LibreTexts: The Kok Cycle and Oxygen Evolving Complex of Photosystem II – https://bio.libretexts.org/Bookshelves/Biochemistry/Fundamentals_of_Biochemistry_(Jakubowski_and_Flatt)/02:_Unit_II-_Bioenergetics_and_Metabolism/20:_Photosynthesis_and_Carbohydrate_Synthesis_in_Plants/20.02:_The_Kok_Cycle_and_Oxygen_Evolving_Complex_of_Photosystem_II
• Govindjee (Univ. of Illinois): Photosynthesis and the "Z"-scheme – https://www.life.illinois.edu/govindjee/textzsch.htm
• ASM.org: The Great Oxidation Event: How Cyanobacteria Changed Life – https://asm.org/articles/2022/february/the-great-oxidation-event-how-cyanobacteria-change

← All articles